diff options
author | Hans de Goede <hdegoede@redhat.com> | 2018-06-02 22:16:38 +0200 |
---|---|---|
committer | Zhang Rui <rui.zhang@intel.com> | 2018-07-26 16:02:48 +0800 |
commit | 54c5848c214922e2c3bf07df6c5789767ff93893 (patch) | |
tree | a9ad2df6bc9335fd23ff33fa1538fca0a4e4806c /Documentation/devicetree/bindings/thermal/hisilicon-thermal.txt | |
parent | d72e90f33aa4709ebecc5005562f52335e106a60 (diff) | |
download | linux-54c5848c214922e2c3bf07df6c5789767ff93893.tar.gz linux-54c5848c214922e2c3bf07df6c5789767ff93893.tar.bz2 linux-54c5848c214922e2c3bf07df6c5789767ff93893.zip |
Thermal: Intel SoC DTS: Translate IO-APIC GSI number to linux irq number
The Intel SoC DTS uses a hardcoded GSI number, before this commit
it was passing it to request_irq as if it were a linux irq number,
but there is no 1:1 mapping so in essence it was requesting a
random interrupt.
Besides this causing the DTS driver to not actually get an interrupt
if the thermal thresholds are exceeded this also is causing an
interrupt conflict on some devices since the linux irq 86 which is
being requested is already in use, leading to oopses like this:
genirq: Flags mismatch irq 86. 00002001 (soc_dts) vs. 00000083 (volume_down)
CPU: 0 PID: 601 Comm: systemd-udevd Tainted: G C OE 4.17.0-rc6+ #45
Hardware name: Insyde i86/Type2 - Board Product Name, BIOS CHUWI.D86JLBNR03 01/14/2015
Call Trace:
dump_stack+0x5c/0x80
__setup_irq.cold.50+0x4e/0xac
? request_threaded_irq+0xad/0x160
request_threaded_irq+0xf5/0x160
? 0xffffffffc0a93000
intel_soc_thermal_init+0x74/0x1000 [intel_soc_dts_thermal]
This commit makes the intel_soc_dts_thermal.c code call
acpi_register_gsi() to translate the hardcoded IO-APIC GSI number (86)
to a linux irq, so that the dts code uses the right interrupt and we
no longer get an oops about an irq conflict.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Zhang Rui <rui.zhang@intel.com>
Diffstat (limited to 'Documentation/devicetree/bindings/thermal/hisilicon-thermal.txt')
0 files changed, 0 insertions, 0 deletions